Jonathan Edwards 10/09/18
Thoughts on my new framing
- Is this retrofitted onto old or need from start?
- Need to make it clear (not macro langauge on outside like Appletalk), but it’s more of a new langauge to build from ground up.
- It’s good to distinguish from Smalltalk. As long as you have Dan Engels sitting next to you, you can do anything you want with your application on the fly. The problem of course is that normal people can’t do this. Very very complicated. My goal is that level of customizability but with much better learning curve. Also not insulated and cult-ish. Modular and inter-operable.
- Distinguish from his vision: he’s pushing very hard on the learning curve. Willing to sacrifice a lot for that, such as complexity, performance, scale, etc, he doesn’t care. He things more of building rather than customization, but they are really the same thing because you always start with something, never from Scratch. There’s a very important social component. It should be like the web. You should very easily get at anything in this world and use a piece of it and share it. And multi-user app, social computer.
- And in terms of approach, he’s into the semantics of this thing as really important.
- His thoughts on my “puzzle piece metaphor” and now “story boarding”… he thinks of it similarly. It’s not going to happen in one giant leap. Have to solve one at a time. Focus on identifying hard problems, make some progress on that, sketching in different pieces…
- It comes out of the mist sometimes: this is a problem I can state. This is a subproblem I can dig into.
- London friends…?
- Alan blackwell and people (person he knows)
- Ask for ppig people after he goes
- Roly intro
- Tomas intro
- Schedule podcast!
- feeling tired of little UI vignettes of an idea, little fragments
- I don’t think other people learn that much from them. A bit but not a lot
- He wants to build something much concrete and understandable to others. Others can dig into
- Wants to build something solid, not thrown away every 12 months. Can build onto in the future
- Shift modes to define a real complete language with a real definition and formal semantics
- Need the greek letters to communicate novel semantics
- subtext has an effect system that can let you know what side-effects code can have. In a very restricted tree-shaped world, I can have very precise effects, and can surface those and show you where in the code, side-effects may be happening. And maybe I can surface them in a very usable way to prevent bugs.
- it’s hot now: rust ownership types, and algebraic effects, etc
- It’d be nice to put a steak in the ground: this is subtext. Stop waffling. Get a running implementation people can download and play with themselves. Could lead to more engagement
- Putting off UI work for a while
- Tomas is a big F# proponent
- Rust has a good story for getting to WASM because WASM needs garbage collection. Rust was created to rewrite firefox rendering